EMD分解、分形理论和RBF神经网络相结合的轴承智能故障诊断研究  

EMD decomposition,fractal theory and RBF neural network fault diagnosis of bearing intelligence research

作　　者：张梅军[1] 陈灏[1] 曹勤 王闯[1] 

机构地区：[1]中国人民解放军理工大学野战工程学院,江苏南京210007 [2]国信证券有限公司基建办,广东深圳518001

出　　处：《机械》2012年第11期10-15,共6页Machinery

基　　金：国家自然科学基金资助项目(51175511)

摘　　要：准确快速的诊断并解决机械设备中的常用组成部件滚动轴承的故障对机械设备和生产至关重要。为了对滚动轴承进行准确的智能诊断,将EMD分解、分形理论和神经网络有机结合,通过运用EMD对信号进行提取和分解,得到其IMF分量,然后画出各IMF分量的关联积分双对数曲线图并从中得出信号的关联维数,借助关联维数并运用RBF神经网络对轴承的状态进行分类和识别,达到智能诊断的目的。实例分析表明EMD分解、分形理论和RBF神经网络相结合能够有效的减少非线性分量对故障信号的干扰并准确识别滚动轴承的故障类型,证明了三者结合的智能故障诊断有效可行。The diagnosis of accurate and fast solving mechanical equipment and production is very important.In order to improve the veracity of the intelligent diagnosis of rolling bearing,this paper will use the organic combination of EMD decomposition,fractal theory and the neural network,get the IMF component,and then draw the IMF component connection points bi-logarithm charts and draw the signal correlation dimension.Use the correlation dimension and RBF neural network to bearing state of the classification and recognition,to achieve the purpose of intelligent diagnosis.Example analysis shows that EMD decomposition,fractal theary and RBF neural network can effectively reduce the nonlinear component of fault signal interference,and accurately identify the rolling bearing fault type.The combination of the three proved effective and feasible intelligent fault diagnosis.

关 键 词：故障诊断 滚动轴承 EMD分解 关联维数 RBF神经网络 

分 类 号：TH17[机械工程—机械制造及自动化]